Nvidia Pushes Open Models Into Robotics Amid Job and Software Fears

Nvidia's latest announcements at Computex in Taiwan highlight the company's expanding role in developing AI systems that interact with the physical world. CEO Jensen Huang outlined several new platforms aimed at robotics and autonomous systems while addressing concerns about how these technologies might affect existing software businesses.

Huang argued that the rise of agentic AI, systems capable of completing tasks with limited human input, would increase demand for software rather than eliminate it. He noted that agents would rely on more tools than before, requiring software interfaces designed specifically for machine use. This view contrasts with recent market concerns that AI could disrupt companies like Salesforce and Workday, which have seen stock declines amid broader questions about software viability.

Alongside these comments, Nvidia introduced the Isaac Gr00t reference design, a humanoid robot platform combining a Unitree H2 chassis with tactile five-fingered hands and Jetson Thor computing hardware. The system provides 31 degrees of freedom across the body and supports whole-body control with significant torque capacity. Researchers can use the accompanying Gr00t software and models to accelerate development of robots that perform physical tasks, with support also planned for lower-cost models like the Unitree G1.

Nvidia also released Cosmos 3, an open AI world model trained on 20 trillion tokens of multimodal data, including images, videos, audio, text, and action sequences from humans and machines. The model generates action data such as joint angles and trajectories, allowing developers to simulate rare or hazardous scenarios for training robots and vehicles. Partners including Agile Robots and Black Forest Labs are participating in an initial coalition to adapt the system across hardware platforms.

These releases build on Nvidia's earlier Gr00t N1 model from March and reflect a deliberate shift toward foundational tools that others can customize. The open approach aims to align future iterations more closely with industry needs while reducing barriers for companies building physical AI applications.

Tech investor Bill Gurley has placed such developments in historical context. Speaking on the All-In Podcast, he compared current fears of AI-driven job losses to warnings during the Industrial Revolution, including those in Pope Leo XIII's 1891 encyclical on labor and inequality. Gurley suggested that similar concerns proved overstated as economies adapted and new roles emerged. Recent comments from Pope Leo XIV on AI and employment indicate the debate continues in policy and institutional circles.

The announcements underscore Nvidia's strategy of supplying both hardware and models that enable broader experimentation in robotics and automation. How these tools affect employment patterns and software markets will depend on adoption rates and the pace at which organizations integrate agentic systems into existing workflows.